Module:Delink/sandbox

-- This module de-links most wikitext.

require('Module:No globals')

local p = {}

local getArgs

local function delinkReversePipeTrick(s)
    if s:match("^%[%[|.*[|\n]") or s == "[[|]]" then -- Check for newlines or multiple pipes.
        return s
    else
        return s:match("%[%[|(.*)%]%]")
    end
end

local function delinkPipeTrick(s)
	-- s the tile area, without bracket nor pipe
    -- We need to deal with colons, brackets, and commas, per [[Help:Pipe trick]].
    
    -- First, remove the text before the first colon, if any.
    s = s:gsub("^(.-:)", "")
    
    -- Next up, brackets and commas.
    if s:match("%(.-%)$") then -- Brackets trump commas.
        s = s:match("(.-) ?%(.-%)$")
    elseif s:match(",") then -- If there are no brackets, display only the text before the first comma.
        s = s:match("(.-),.*$")
    end
    return s
end

local function delinkWikilink(s)
	-- s is a string starting with "[[" and ending with "]]". It does not contain any other "]]" strings.
	
	-- Deal with nested links
	local nested = "[[" .. s:sub(3):gsub("%[%[.-%]%]$", delinkWikilink)
    if nested ~= s then
    	return nested
    end
    
    -- Deal with the reverse pipe trick.
    if s:match("^%[%[|") then
        return delinkReversePipeTrick(s)
    end
    
    -- Check for bad titles. To do this we need to match the
    -- title area of the link, i.e. the part before any pipes.
    local titlearea, display = s:match("^%[%[(.-)|(.*)%]%]$")
    titlearea = titlearea or s:sub(3, -3)
    
    titlearea = mw.uri.decode(titlearea, "PATH") -- decode percent-encoded entities. Leave underscores and plus signs.
    titlearea = mw.text.decode(titlearea, true) -- decode HTML entities.
        
    local temptitlearea, fragment = titlearea:match("^(.-)#(.*)$")
    temptitlearea = temptitlearea or titlearea
    fragment = fragment or ""
    
    -- Check for bad characters.
	if not mw.ustring.isutf8(titlearea)
		or temptitlearea:match("[%[%]<>|{}%z%c\n]") 
		or temptitlearea:match("%%%x%x")
		or temptitlearea:match("&..-;") 
		or fragment:match("[%[%]{}%c\n]") 
	then
    	return s
    end
    
    
    -- Check for categories, interwikis, and files.
    local colonprefix = titlearea:match("^(.-):") or "" -- Get the text before the first colon.
    local ns = mw.site.namespaces[colonprefix] -- see if this is a known namespace
    if mw.language.isKnownLanguageTag(colonprefix)
    or ( ns and ( ns.canonicalName == "File" or ns.canonicalName == "Category" ) ) then
    	return ""
    end
    
    -- Remove the colon if the link is using the [[Help:Colon trick]].
    if titlearea:sub(1, 1) == ":" then
        titlearea  = titlearea:sub(2)
    end
    
    -- Deal with links using the [[Help:Pipe trick]].
    if display == "" then
        return delinkPipeTrick(titlearea )
    end
    
    -- match the display area of the wikilink
    if not display then -- match if we're dealing with a piped link.
        -- Remove new lines from the display of multiline piped links,
        -- where the pipe is before the first new line.
        titlearea = titlearea:gsub("\n", "")
    end

    return display or titlearea 
end

local function delinkURL(s)
    -- Assume we have already delinked internal wikilinks, and that
    -- we have been passed some text between two square brackets [foo].
    
    -- If the text contains a line break it is not formatted as a URL, regardless of other content.
    if s:match("\n") then
        return s
    end
    
    -- Check if the text has a valid URL prefix and at least one valid URL character.
    local valid_url_prefixes = {"//", "http://", "https://", "ftp://", "gopher://", "mailto:", "news:", "irc://"} 
    local url_prefix
    for i,v in ipairs(valid_url_prefixes) do
        if s:find( '^%[' .. v ..'[^"%s].*%]' ) then
            url_prefix = v
            break
        end
    end
    
    -- Get display text
    if not url_prefix then
    	-- Deal with nested links or send back original string.
        return "[" .. mw.ustring.gsub( mw.ustring.sub(s, 2), "%[.-%]", delinkURL )
    end
    s = s:match("^%[" .. url_prefix .. "(.*)%]") -- Grab all of the text after the URL prefix and before the final square bracket.
    s = s:match('^.-(["<> [].*)') or "" -- Grab all of the text after the first URL separator character ("<> ).
    s = s:match("^%s*(%S.*)$") or "" -- If the separating character was a space, trim it off.
    
    local s_decoded = mw.text.decode(s, true)
    if s_decoded:match("%c") then
        return s
    else    
        return s_decoded
    end
end


function p._delink(args)
    local text = args[1] or ""
    if args.refs == "yes" then
        -- Remove any [[Help:Strip markers]] representing ref tags. In most situations 
        -- this is not a good idea - only use it if you know what you are doing!
        text = mw.text.unstrip(text)
    end
    if not (args.comments == "no") then
        text = text:gsub("<!%-%-.-%-%->", "") -- Remove html comments.
    end
    if not (args.wikilinks == "no") then
        text = text:gsub("%[%[.-%]%]", delinkWikilink) -- De-link wikilinks.
    end
    if not (args.urls == "no") then
        text = text:gsub("%[.-%]", delinkURL) -- De-link URLs.
    end
    if not (args.whitespace == "no") then
        -- Replace single new lines with a single space, but leave double new lines
        -- and new lines only containing spaces or tabs before a second new line.
        text = text:gsub("([^\n \t][ \t]*)\n([ \t]*[^\n \t])", "%1 %2")
        --text = text:gsub("\n\t+", "\n") -- Remove leading tab.
        text = text:gsub("[ \t]+", " ") -- Remove extra tabs and spaces.
    end
    if not (args.nowiki == "no") then
    	text = mw.text.unstripNoWiki(text)
    end
    return text
end

function p.delink(frame)
    if not getArgs then
    	getArgs = require('Module:Arguments').getArgs
    end
    return p._delink(getArgs(frame, {wrappers = 'Template:Delink'}))
end

return p